ARTICLE

Received 18 Sep 2013 | Accepted 11 Feb 2014 | Published 11 Mar 2014 DOI: 10.1038/ncomms4430 OPEN HACE1-dependent protein degradation provides cardiac protection in response to haemodynamic stress

Liyong Zhang1,2, Xin Chen1,2, Parveen Sharma2, Mark Moon1,2, Alex D. Sheftel1, Fayez Dawood1,2, Mai P. Nghiem2, Jun Wu2, Ren-Ke Li2, Anthony O. Gramolini2,3, Poul H. Sorensen4, Josef M. Penninger5, John H. Brumell6,7,8 & Peter P. Liu1,2,3,7

The HECT E3 ubiquitin ligase HACE1 is a tumour suppressor known to regulate Rac1 activity under stress conditions. HACE1 is increased in the serum of patients with heart failure. Here we show that HACE1 protects the heart under pressure stress by controlling protein degradation. Hace1 deficiency in mice results in accelerated heart failure and increased mortality under haemodynamic stress. Hearts from Hace1 À / À mice display abnormal cardiac hypertrophy, left ventricular dysfunction, accumulation of LC3, p62 and ubiquitinated proteins enriched for cytoskeletal species, indicating impaired autophagy. Our data suggest that HACE1 mediates p62-dependent selective autophagic turnover of ubiquitinated proteins by its ankyrin repeat domain through protein–protein interaction, which is independent of its E3 ligase activity. This would classify HACE1 as a dual-function E3 ligase. Our finding that HACE1 has a protective function in the heart in response to haemodynamic stress suggests that HACE1 may be a potential diagnostic and therapeutic target for heart disease.

1 University of Ottawa Heart Institute, 40 Ruskin Street, Ottawa, Ontario, Canada K1Y 4W7. 2 Heart and Stroke/Richard Lewar Centre of Excellent for Cardiovascular Research, University of Toronto and Toronto General Research Institute, University Health Network, Toronto, Ontario, Canada M5G 2C4. 3 Department of Physiology, University of Toronto, Toronto, Ontario, Canada M5S 1A8. 4 Department of Molecular Oncology, BC Cancer Research Center, University of British Columbia, Vancouver, British Columbia, Canada V5Z 1L3. 5 Institute of Molecular Biotechnology of the Austrian Academy of Sciences, Dr Bohrgasse 3, A-1030 Vienna, Austria. 6 Department of Molecular Genetics, University of Toronto, Toronto, Ontario, Canada M5S 1A8. 7 Institute of Medical Science, University of Toronto, Toronto, Ontario, Canada M5S 1A8. 8 Cell Biology Program, Hospital for Sick Children, Toronto, Ontario, Canada M5G 1 Â 8. Correspondence and requests for materials should be addressed to P.P.L. (email: [email protected]).

NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications 1 & 2014 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430

eart disease is a major cause of death in the developed investigate Hace1’s role in the progression of heart failure, we world, and protection against heart failure is a major subjected Hace1 À / À and Hace1 þ / þ mice to hypertrophic heart Hchallenge1. Animal models have shown that inappropriate failure induced by sTAC20. Sham-operated mice were used as responses of the heart to sustained haemodynamic stress can lead controls. Starting from the second day after surgery, noticeable to heart failure2–4. Protein degradation pathways, including the signs of circulatory failure, including lethargy, impaired mobility, proteasome and autophagy, play a protective role in the response diminished appetite and peripheral oedema were observed in to stress, and their dysregulation appears to underline multiple mice subjected to sTAC. Echocardiographic measurements forms of myocardial disease5–10. In post-mitotic cells such revealed no cardiac physiological functional differences between as cardiomyocytes, turnover of proteins and organelles is Hace1 À / À , Hace1 þ / þ or sham-operated mice at baseline particularly important for cellular protein quality control and (Fig. 1c–e). However, pressure overload induced by sTAC maintenance. Two major pathways within the cell that regulate triggers dramatic decreases in left ventricular (LV) function as proteolysis are the ubiquitin (Ub)–proteasome system (UPS) and measured by echocardiography at days 2 and 4 post operation the autophagy lysosomal pathway6. Whereas the UPS is mainly (Fig. 1c–e). More importantly, these signs of LV dysfunction were involved in the rapid degradation of signalling proteins with more notable in the Hace1 À / À sTAC-treated mice as shown by a short half-lives, autophagy can selectively remove more significantly decreased percentage of LV fractional shortening persistent protein aggregates and damaged organelles5. Both and severely dilated LV, which was indicated by a significant proteasome-mediated degradation and selective autophagy increase in both LV end-diastolic dimension and LV end-systolic require ubiquitination of the cargo, which is then recognized by dimension compared with wild-type (WT) mice with the same Ub receptors directing it to either 26S proteasomes or lysosome treatment (Fig. 1c–e). Moreover, The LV dysfunction resulted in for degradation6. HACE1, as a novel stress inducible Ub E3 significantly higher mortality rates in Hace1 À / À mice following protein ligase, has potential roles in both proteolytic systems. sTAC; 40% (8/20) of Hace1 À / À mice died between day 1 and day We and others previously reported that HACE1 is involved 4 post sTAC, compared with only 16% (3/19) of the Hace1 þ / þ in proteasome-mediated protein degradation11–17. However, mice (log-rank w2-test, Po0.001; Fig. 1f). This was accompanied HACE1’s involvement in autophagy-mediated protein degra- by a significant increase in cardiac mass as measured by growing dation has never been explored. heart weight/body weight ratio in Hace1 À / À sTAC mice Increases in autophagic flux have been documented in virtually compared with Hace1 þ / þ sTAC mice (Fig. 1g). An increase in all forms of human cardiovascular disease, including ventricular myocyte cross-sectional area, as revealed by haematoxylin and hypertrophy and heart failure18,19. This is indicative of a eosin staining of left ventricle cross-sections, confirmed that the requirement for autophagy in the maintenance of cellular increase in heart weight/body weight ratio in Hace1 À / À sTAC homoeostasis during ventricular hypertrophic remodelling. mice was caused by myocyte enlargement (Fig. 1i). Consistent Although HACE1 is highly expressed in the heart, nothing is with severe heart failure, Hace1 À / À sTAC mice also had a known about its function in the heart. The emerging role of other significantly higher lung weight/body weight ratio, accompanied E3 ligase and Ub-binding proteins in selective autophagy suggests by pulmonary oedema, compared with Hace1 þ / þ sTAC mice that HACE1 may also participate in this process. (Fig. 1h). Expression of atrial natriuretic peptide precursor A, a Here we demonstrate for the first time that HACE1 is essential biomarker of cardiac stress and hypertrophy21–23, was also for the completion of selective autophagic flux. Hace1 deficiency elevated in Hace1 À / À sTAC hearts (Fig. 1j). In addition, in our results in increased accumulation of ubiquitinated protein initial pilot studies of HACE1 protein level changes in human aggregates in stressed myocardium due to impaired autophagy clinical plasma samples with heart failure, we also observed some and thus accelerates cardiac dysfunction. degree of correlation between HACE1 and B-type natriuretic peptide (Supplementary Fig. 1). However, the data only explain 2 Results 19.5% of the relationship (as per R value), and thus HACE1 and HACE1 is upregulated in human heart failure. To clarify the B-type natriuretic peptide do ultimately reflect different biological processes. These findings indicate that, in the absence of Hace1, role of HACE1 in the development of heart failure, we performed in silico GEO profiles’ analysis of HACE1 gene expression related to haemodynamic stress induced by sTAC triggers rapid decompensated cardiac hypertrophic remodelling, progressing human heart disease. Our search indicated that HACE1 gene expression is upregulated in dilated cardiomyopathy (DCM) to heart failure and increased mortality in these mice. compared with non-failing hearts (GEO accession code GDS2206). Accumulation of protein aggregates in Hace1 À / À sTAC To confirm this, we examined HACE1 messenger RNA expression myocardium. To explore Hace1’s role in controlling cardiac in a panel of heart tissue biopsies from heart failure patients as well as non-failing heart controls. Significantly, increased HACE1 gene remodelling at the cellular level, LV sections of sTAC mice were subjected to transmission electron microscopy (TEM) examination. expression was detected in cardiac tissues of heart failure patients À / À compared with non-failing hearts; there was a 2.5-fold increase in Ultrastructural analysis of Hace1 sTAC hearts showed a dis- organized sarcomere structure and significantly more structures HACE1 gene expression in ischaemic cardiomyopathy and more than a 3-fold increase in DCM (Fig. 1a). In addition, markedly resembling autophagic vacuoles (Supplementary Fig. 2a). It is noteworthy that these cellular phenotypes are similar to those elevated Hace1 gene expression was detected in a murine model of heart failure induced by severe transverse aortic constriction previously observed in Atg5-deficient hearts, which are defective in protein/organelle turnover via autophagy5. We also observed fewer (sTAC) compared with a sham-operated control group at 4 days À / À post operation (unpaired two-tailed Student’s t-test, P ¼ 0.0004; high-density autolysosome structures in Hace1 sTAC hearts when compared with Hace1 þ / þ sTAC hearts (Supplementary Fig. 1b). The robust induction of HACE1 expression in the heart under pathological stress conditions suggests that HACE1 parti- Fig. 2a,b). In contrast, we observed enlarged autophagosome-like vacuole structures, in some cases containing intact mitochondria, cipates in stress-induced cardiac hypertrophic remodelling. suggesting a disruption in the completion of the autophagy pathway (Supplementary Fig. 2a). Ventricular sections from sham-treated Loss of Hace1 accelerates load-induced heart failure in mice. Hace1 þ / þ and Hace1 À / À mice appeared normal (Supplementary Surgical models of cardiac pressure overload in mice have pro- Fig. 2c,d). Thus, Hace1 deficiency results in extensive cellular vided a platform for studying load-induced heart failure. To defects in the myocardium during haemodynamic stress.

2 NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications & 2014 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430 ARTICLE

abP<0.0001 P=0.0004 5 P<0.01 P<0.0001 5

4 4 mRNA mRNA 3 3 HACE1 2 Hace1 2

1 1 Relative Relative 0 0 ControlICM DCM Sham sTAC c 60 d 4 50 P<0.001 40 3 P<0.001

30 P<0.001 P<0.001 2 20

LVESD (mm) 1 10 Fractional shortening (%) 0 0 01234 01234 Time post surgery (days) Time post surgery (days) e 5.0 f 100

4.5 80 P<0.01 P<0.05 P<0.01 Hace1+/+ sham 4.0 –/– LVEDD (mm) 60 Hace1 sham Percent survival Hace1+/+ sTAC Hace1–/– sTAC 3.5 40 0 1 234 1 234 Time post surgery (days) Time post surgery (days)

Hace1+/+ Hace1–/– g P<0.05 i 12 P<0.001

8 NS Sham

HW/BW 4 P<0.001 0 sTAC

h P<0.05 20 j 150 P<0.001 15 P<0.001

100 +/+ 10 NS Hace1 mRNA LW/BW Hace1–/– 5 50 P<0.05 Anp 0 C 0 Sham sTAC WT sham KO sham WT sTAC KO sTA

Figure 1 | Hace1 deficiency accelerates pressure overload-induced heart failure. (a) HACE1 mRNA expression is markedly increased in ischaemic and dilated cardiomyopathy (ICM and DCM) heart failure (HF) patient heart tissue compared with non-failing hearts (NF). Expression levels were normalized to HPRT and GAPDH mRNAs. n ¼ 15 per group. Control, non-failing heart. (b) Elevated Hace1 mRNA expression in sTAC mouse hearts compared with sham controls, expression levels were normalized to Hprt and Gapdh mRNAs. n ¼ 6–9 per group, P ¼ 0.004 (unpaired two-tailed Student’s t-tests). (c–j) Hace1 À / À and WT mice were subjected to sTAC or sham surgery and analysed 4 days after the operation. (c–e) Echocardiographic measurement of (c) LV fractional shortening (FS ¼ (LVEDD À LVESD)/LVEDD Â 100%); (d) LV end-diastolic dimension (LVEDD) (mm) and (e) LV end-systolic dimension (LVESD) (mm) are shown. n ¼ 6–8 for each group. (f) Mortality rate of WT and Hace1 À / À mice after sTAC is shown by Kaplan–Meier plots as the survival proportion of each cohort of mice. Survival incidence in mice over the indicated follow-up period is shown as the ratio of the number of mice survived/total mice analysed for each group. This represents 60% (12/20) for Hace1 À / À sTAC mice, 84% (16/19) for Hace1 þ / þ sTAC mice and 100% (12/12) for both Hace1 À / À and Hace1 þ / þ sham mice on day 4; P value as indicated (log-rank w2-test). (g) Heart weight/body weight ratios (HW/BW) and (h) lung weight/body weight ratio (LW/BW) of WT and Hace1 À / À mice 4 days after sTAC or sham surgery. n ¼ 15–21 for each group. (i) Representative haematoxylin and eosin staining transverse heart cross-sections showing myocyte cross-sectional area (scale bars, 20 mm). (j) Real-time RT–PCR analyses reveal a significant increase in the expression of the atrial natriuretic peptide gene (Nppa), a clinical biomarker for heart hypertrophy, in Hace1 À / À sTAC heart; expression levels were normalized to Hprt and Gapdh mRNAs. n ¼ 6–9 per group. In all panels, error bars represent s.e.m., one-way analysis of variance were used to calculate P-values.

NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications 3 & 2014 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430

One of the impacts of impaired protein degradation pathways the LC3-II forms of the protein accumulated to a significantly is the accumulation of ubiquitinated proteins5,24. Immuno- higher level in Hace1 À / À sTAC hearts when compared with histochemical staining of left ventricle cross-sections of sTAC Hace1 þ / þ controls. Enhanced LC3 accumulation was also mice revealed a significantly increased accumulation of observed in Hace1 À / À /LC3Tg sTAC myocardium, which stably ubiquitinated proteins in Hace1 À / À myocardium (Fig. 2a). expresses green fluorescence protein (GFP)-LC3 under the This was further confirmed by western blot analysis (Fig. 2d,e). b-actin (CAG) promoter26 (Fig. 2h). Moreover, no significant The Ub and microtubule-associated protein 1 light chain 3B increase in p62 or Lc3b gene expression was detected in Hace1 À / À (LC3II) binding protein p62 (also known as sequestosome 1), sTAC heart (Supplementary Fig. 3), suggesting that the increased which is known to be degraded by autophagy, also accumulates p62 and LC3 accumulation were due to impaired degradation by during impairment of autophagy24. We observed by autophagy at the post-transcriptional level. immunohistochemical analysis (Fig. 2b) and western blot It is possible that defects in proteasomal activity were (Fig. 2d,f) increased levels of p62 protein in the Hace1 À / À responsible for the ubiquitinated protein accumulation in sTAC heart tissue. This suggests that autophagic degradation of sTAC Hace1 À / À myocardium. However, heart lysates from p62 was blocked in the absence of Hace1, resulting in the sTAC-treated Hace1 À / À mice displayed elevated 20S protea- increased cellular accumulation of p62. some levels compared with Hace1 þ / þ hearts (Fig. 2i). This Next, we assessed autophagy by examining microtubule- may reflect a compensating mechanism to overcome the associated protein LC3, an autophagy protein that is present in impairment of autophagy. Indeed, elevated cardiac proteasomal the cytosol (LC3-I form) until it becomes targeted to nascent activity accompanies autophagy deficiency on deletion of autophagosomes via conjugation to phosphatidylethanoloamine Atg5 (ref. 5). (LC3-II form)25. Levels of LC3 were first examined by western Thus, Hace1 deficiency impaired the clearance of ubiquitinated blot (Fig. 2c,g). In Hace1 À / À sTAC hearts, we observed a proteins in the myocardium and is associated with the significant increase in total LC3 protein levels. Both LC3-I and acceleration of heart failure. In addition to its role in protein

+/+ –/– a Hace1 Hace1 d Hace1+/+ Hace1–/– kDa Sham sTAC Sham sTAC Sham 10μm10μm 50 proteins +

Ub 30 proteins sTAC

10μm10μm + Ub b 15 Sham 10μm10μm 50 p62 p62 30 Gapdh

sTAC e P<0.05 f P<0.001 10μm10μm 15 2 Hace1+/+ 10 c Hace1–/– 1 proteins/

Gapdh 5 + Sham

μ μ p62/Gapdh 10 m10m Ub 0 0 Sham sTAC Sham sTAC

GFP-LC3II g Hace1+/+ Hace1–/–

sTAC Sham sTAC Sham sTAC 10μm10μm kDa LC3I i 2.5 P<0.05 15 LC3II Gapdh 2.0 30

1.5 h Hace1+/+ P<0.0001 2 4 Hace1–/– P<0.0001 P<0.05 activity 1.0 NS 3 0.5 1 2 Relative proteasome Relative 1 LC3I/Gapdh 0.0 LC3II/Gapdh 0 0 Sham sTAC Sham sTAC

WT sham KO sham WT sTAC KO sTAC

Figure 2 | Accumulation of p62 LC3 and Ub þ protein in Hace1 À / À sTAC hearts. Representative immunohistochemical staining for (a) ubiquitinated protein (brown); (b) p62 (brown) and (c) GFP-LC3 (brown) in paraffin sections of heart revealing increased accumulation of p62, ubiquitinated protein and LC3 in Hace1 À / À sTAC hearts. Hearts from three mice in each group were analysed (scale bars, 10 mm). Representative immunoblot and quantification for (d,e) Ub-conjugated proteins, (d,f) p62 and (g,h) LC3 in heart extracts (n ¼ 3 per group), Gapdh were used as a loading control. (i) Increased Ub proteasome activity in Hace1 À / À sTAC hearts as measured by 20S proteasome activity assay (n ¼ 4–6 per group). In all panels, error bars represent s.e.m., P value between KO and WT sTAC as indicated (one-way analysis of variance).

4 NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications & 2014 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430 ARTICLE degradation by the proteasome11–15, our data indicate that Besides the HECT domain, the other known domain of HACE1 is also required for efficient functioning of the autophagy HACE1 is the six ankyrin repeats (ANK) at its amino terminal. pathway in the heart during haemodynamic stress. However, the function of the ANK domain in HACE1 is entirely unknown. On the basis of our finding that HACE1 promotes ubiquitinated protein aggregate clearance independently of its E3 HACE1 controls autophagic clearance of protein aggregates. ligase activity, we hypothesize that HACE1 functions as an To better understand HACE1’s function in the clearance of adaptor protein, binding to ubiquitinated proteins and delivering ubiquitinated proteins under stress conditions, we examined its them for autophagic degradation via HACE1’s ANK domain. role in protein aggregate clearance in both primary neonatal To test this hypothesis, three GFP-tagged HACE1 plasmids, cardiomyocytes (NCMs) and embryonic fibroblasts (MEFs) WT HACE1-GFP, HACE1DHECT-mGFP and HACE1DANK- derived from Hace1 À / À and Hace1 þ / þ control mice. Ubiqui- mGFP, were generated. The two mutant HACE1 constructs were tinated protein aggregate formation was induced by treatment generated by DNA synthesis with deletion of either the carboxy- with puromycin, an inhibitor of protein translation that induces terminal HECT domain (DHECT) or the N-terminal ANK repeat premature polypeptide chain termination and the subsequent (DANK; Fig. 4a). After transient transfection of the plasmid into accumulation of misfolded proteins, which can be detected with a Hace1 À / À MEF, MEF were subjected to a protein aggregate monoclonal antibody specific to ubiquitinated proteins27. clearance assay. Similar to WT HACE1 (Fig. 4b), HACE1D HECT Immunofluorescence staining shows HACE1 localized to localized to ubiquitinated protein aggregates in puromycin- ubiquitinated protein aggregates in puromycin-treated MEF treated MEF. Moreover, similar to WT HACE1, the majority of (Fig. 3a). This suggests that HACE1 may act locally at the HACE1D HECT cells were able to clear aggregates by 8 h after aggregates to promote protein turnover. To test this hypothesis, puromycin removal (Fig. 4c). In contrast, HACE1DANK did not we subjected WT and Hace1 À / À primary NCMs to an aggregate localize with ubiquitinated protein aggregates and lost the ability clearance assay27,28, where cells were pulsed with puromycin for to clear aggregates after puromycin removal (Fig. 4d). Therefore, 4 h to induce protein aggregate formation. The treatment was HACE1 is required for efficient clearance of protein aggregates then removed by extensive washing, followed by an 8-h chase to and this is independent of its E3 ligase activity. Instead, HACE1 allow for aggregate clearance. The majority of WT NCM was able may participate in the delivery of ubiquitinated protein aggregates to clear aggregates by 8 h after puromycin removal (Fig. 3b). In to the autophagic lysosomal pathway for degradation via protein– contrast, aggregate clearance was almost completely blocked in protein interactions mediated by its ANK domain. Hace1 À / À NCM. To evaluate whether HACE1’s E3 ligase activity is required for HACE1 partners with p62 in protein aggregate clearance. The the clearance of protein aggregates, we next compared the effects adaptor protein p62 has the ability to bind to both ubiquitinated of WT and mutant HACE1 expression on rescue of aggregate proteins and LC3II (ref. 30). In the Hace1 À / À sTAC heart, we clearance. Hace1 À / À MEF stably transfected with either WT observed a significant increase in p62-positive puncta (Fig. 2d). HACE1 complementary DNA (WT-HACE1) or a C876S point To determine whether HACE1 co-localized with p62, Hace1 À / À mutated form of HACE1 (C876S) were used for the rescue studies. NCM transiently expressing HACE1-GFP were subjected to p62 The point mutation of amino acid 876 from cysteine to serine is immunofluorescent staining. Under control conditions, HACE1 known to abolish E3 ligase activity of HACE1 (refs 11–15). As expressed uniformly across the cytoplasm, mostly not localizing expected, the defects in aggregate clearance in Hace1 À / À MEF with p62. However, on puromycin treatment, both HACE1 and were rescued by WT HACE1. Surprisingly, HACE1 (C876S) p62 relocated to the perinuclear region and, more importantly, rescued the protein aggregate clearance defect as efficiently as WT co-localized (Fig. 5a). This indicates that HACE1 may control HACE1 (Fig. 3c,d). protein aggregate degradation via regulation of p62. To further address whether HACE1 partners with p62 in protein aggregate clearance, we depleted p62 by short interfering RNA (siRNA) in ANK domain in HACE1 is required for removal of protein Hace1 þ / þ and Hace1 À / À MEFs (Supplementary Fig. 5a) and aggregates. To date, the only known HACE1 E3 ligase substrate subjected cells to a protein aggregate clearance assay. Knockdown is Rac1. It has been shown that HACE1 controls Rac1 activity of p62 partially restored ubiquitinated protein aggregate degra- through binding and ubiquitinating the GTP-bound form of dation in Hace1 À / À MEF (Supplementary Fig. 5b), suggesting Rac1, leading to its proteasomal degradation13–15. Transgenic p62 could be a downstream target of HACE1. It has been mice overexpressing Rac1 exhibit the phenotype of cardiac reported that p62 promotes the formation of polyubiquitinated hypertrophy and DCM29. To test the possibility that upregulated protein aggregates to allow clearance via autophagy24,30,31. Rac1 may contribute to the heart failure phenotype in Hace1 À / À Therefore, in HACE1 knock-out cells, where autophagy is sTAC mice, we subjected sTAC hearts to Rac1 immunoblotting. impaired, p62 knockdown would inhibit aggregate formation No significant changes in Rac1 protein levels were detected but presumably not total cellular ubiquitination and toxicity. To between Hace1 À / À and WT groups (Supplementary Fig 4a). test whether E3 Ub ligase activity is involved in HACE1’s Although we cannot exclude the possibility that Rac1 may interaction with p62, WT, Hace1 À / À MEFs and Hace1 À / À participate in the cardiac remodelling process under stress, at MEF stably transfected with either WT HACE1 (WT-HACE1)or least in the particular experimental settings that we have tested, a C876S-mutant form HACE1(C876S) were treated with MG132, Rac1 does not appear to be regulated by Hace1. A report by Zhao a specific, potent and cell permeable proteasome inhibitor that et al.17 indicated that HACE1 is a novel repressor of retinoic acid reduces the degradation of Ub-conjugated proteins in receptor (RAR) transcription, a role that is independent from its mammalian cells32–35, for 4 h to induce ubiquitinated protein E3 Ub ligase activity. This suggests that HACE1 has additional aggregates formation. When p62-immunoprecipitated samples function(s) beyond its role as an E3 Ub ligase. To explore this were subjected to immunobloting with FK2 (Ub-conjugated further, we subjected sTAC hearts to RAR-b immunoblotting. We proteins) antibody, we observed increased Ub-conjugated detected no significant changes in RAR-b protein levels between proteins in Hace1 À / À MEF. Adding either WT or C876S Hace1 À / À and WT groups (Supplementary Fig 4b). This mutant back to Hace1 À / À MEF decreased the pull-down of UB- suggests that RAR-b did not significantly contribute to the conjugated proteins to a similar level as WT (Supplementary Fig heart failure phenotype in the Hace1 À / À sTAC mice. 5c). Our data indicate that HACE1 expression can affect p62

NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications 5 & 2014 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430

a Puromycin b Hace1+/+ Hace1–/–

d P<0.001

Overlay 100 80 60 aggregate + 40 clearance 20 Puromycin HA-HACE1 % Of Ub 0 WT y +WT-E1 CE1- +Empt vector +HA C876S

proteins HAC +

Puromycin + recover Vehicle –/– Ub Hace1

Ub+ proteins

c Hace1–/–

WT +Empty vector +WT-HACE1 +HACE1(C876S) Vehicle proteins

+ Puromycin Ub recover Puromycin + Puromycin

Figure 3 | HACE1 is required for autophagic clearance of Ub þ protein aggregates independent of its E3 ligase activity. (a) Representative confocal microscopy images showing HACE1 (green) localizing to the Ub þ protein aggregates (red) induced by puromycin treatment in HA-HACE1 MEF. Arrows indicate co-localized puncta. Scale bar, 10 mm. (b) Representative confocal micrographs of Hace1 À / À and Hace1 þ / þ NCM treated with puromycin for 4 h (puromycin), puromycin followed by 8 h recover (puromycin þ recover). Vehicle (PBS)-treated cells were used as control. Hace1 deficiency in cardiomyocytes (Hace1 À / À NCM) blocks ubiquitinated protein aggregate degradation. Scale bars, 10 mm. (c,d) WT and Hace1 À / À MEF as well as Hace1 À / À MEF stably transfected with WT HACE1 and C876S-mutant HACE1were subjected to protein aggregates clearance assay by pulsing with puromycin for 4 h to induce protein aggregate formation followed by an 8-h chase to monitor for aggregate clearance. Cells were fixed and stained for ubiquitinated proteins to visualize aggregates. Vehicle (PBS)-treated MEFs were used as control. (c) Representative confocal micrographs of images used for quantification of the efficiency of aggregates clearance in d (percentage of cells that cleared all aggregates by 8 h after puromycin removal). At least 10 images from each group were used for the calculation, error bars represent s.e.m., Po0.001 (one-way analysis of variance). Scale bars, 10 mm. levels independently of its E3 Ub ligase activity. Therefore, (Supplementary Fig. 6). This indicates that the induction phase of HACE1 may serve as an adaptor for autophagy in addition to its autophagy is not affected by Hace1 deficiency. role in protein degradation via the Ub–proteasome pathway. Autophagy is a dynamic process with autophagosome forma- tion being balanced by turnover on delivery to lysosomes. HACE1 is required for the completion of autophagic flux. The Inhibition of lysosome-mediated degradation will block LC3 above findings are consistent with the notion that HACE1 is a degradation and hence increase LC3 accumulation20. Enhanced novel potential regulator of the selective autophagy pathway. LC3 accumulation in Hace1 À / À sTAC heart (Fig. 2h) suggests Next, we tried to determine in which step of autophagy HACE1 that HACE1 deficiency could block autophagic flux. To verify participates. Changes in gene expression from a panel of this, HACE1’s interaction with LC3 was examined in Hace1 À / À / autophagy-related genes were analysed by quantitative PCR in GFP-LC3Tg NCM transiently expressing HACE1-RFP. Blockade sTAC and sham hearts. We found no significant differences in of UPS by proteasome inhibition has been shown to activate expression between WT and Hace1 À / À sTAC hearts in genes autophagy32–35. Exposure of the NCM to MG132 led to the co- related to autophagy initiation such as Atg5, Atg7 and Beclin1 localization of HACE1 with LC3 puncta (Fig. 5b). In addition,

6 NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications & 2014 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430 ARTICLE

a b WT HACE1Ub+ proteins overlay

ANKHECT mGFP WT Vehicle C876 ΔHECT

ΔANK

C876 Puromycin + recover Puromycin

cdHACE1 ΔHECTUb+ proteins Overlay HACE1 ΔANK Ub+ proteins Overlay Vehicle Puromycin + recover Puromycin

Figure 4 | ANK domain in HACE1 is necessary for autophagic removal of Ub þ protein aggregates. (a) Diagrams representing GFP-tagged WT HACE1 and mutant HACE1 (HACE1D HECT and HACE1D ANK) constructs transiently transfected into Hace1 À / À MEF for protein aggregate clearance assay. (b–d) Representative confocal micrographs of (b)WT,(c) DHECT and (d) DANK MEF subjected to protein aggregate clearance assay by pulsing with puromycin for 4 h to induce protein aggregate formation followed by an 8-h chase to monitor for aggregate clearance. Vehicle-treated MEFs were used as control. Scale bars, 10 mm. when we subjected MG132-treated Hace1 À / À /GFP-LC3Tg and treatment with Baf reversed this pattern. In contrast, loss of NCM and control Hace1 þ / þ /GFP-LC3Tg NCM cells to LC3 Hace1 resulted in a significant increase in green/red-positive immunoblot analysis, HACE1 deficiency clearly led to increased autophagosome structures with a reduction in red-only GFP-LC3 accumulation in Hace1 À / À /GFP-LC3Tg NCM com- autolysosome structures in the absence of Baf as shown in pared with WT control (Fig. 5c). This suggests that lysosome- Hace1 À / À NCM. In addition, no significant changes were mediated degradation is impaired in the absence of HACE1. observed in Hace1 À / À NCM co-treated with Baf (Fig. 6a,b). This Indeed, HACE1-GFP-labelled structures also co-localized with indicates that there is a defect in autophagosome–lysosome fusion the lysosomal marker, Lamp1 (Fig. 5d), suggesting that HACE1 in Hace1-deficient cardiomyocytes. To test whether the E3 ligase may be required for autolysosome formation. activity of Hace1, if required for this process, we next tested the To determine whether HACE1 is required for the fusion of tfLC3 reporter in Hace1 À / À MEF stably transfected with empty autophagosomes with lysosomes, we used a tandem tagged vector (MSCV), WT HACE1 (HA-HACE1) and E3 ligase dead mRFP-GFP-LC3 (tfLC3) probe. This probe can be used to detect HACE1 (C876S-HACE1). WT MEF was used as a control. After whether an autophagosome has fused with a lysosome, based on 4 h incubation with MG132, WT MEF had fewer green/red the distinct chemical properties of GFP and mRFP fluorophores. autophagosomes and more red autolysosomes; in contrast, Under neutral pH conditions in the cytosol, both GFP and mRFP significantly increased green/red-positive autophagosome struc- moieties of the probe fluoresce. However, under low pH tures and reduced red-positive-only autolysosome structures were conditions in the lumen of the lysosome, the GFP signal is observed in Hace1 À / À (MSCV) MEF. As expected, the defects quenched, but not the mRFP. By exploiting the difference in the in autophagosome–lysosome fusion in Hace1 À / À MEF were nature of these two fluorescent proteins, autophagic flux can be rescued by WT HACE1. Moreover, E3 ligase mutation in HACE1 measured with this mRFP-GFP-LC3 tandem probe36. (C876S MEF cells) results in similar mRFP-GFP-LC3 expression First, we tested the tfLC3 reporter in WT and Hace1 À / À patterns as WT (Fig. 6c,d). This suggests that HACE1 is required primary NCMs 48 h after transient transfection with tfLC3 for the fusion of autophagosomes with lysosomes independent of plasmid. NCM were treated with MG132 for 8 h to induce its E3 ligase activity. autophagy, as previously described34. In addition, bafilomycin A1 (Baf), an inhibitor of lysosomal acidification, was added to one of HACE1 is required for maintaining protein homoeostasis.To the MG132-treated groups in the last 4 h to inhibit autolysosome identify HACE1-interacting proteins, a combination of co- formation. MG132 is capable of inducing autophagy in WT and immunoprecipitation assay (Co-IP) with mass spectrometry (MS) Hace1 À / À NCM, as measured by increased GFP-LC3-positive analysis was conducted in Hace1 À / À MEF stably expressing an puncta compared with the untreated group (Fig. 6a). However, N-terminal haemagglutinin (HA)-tagged full-length HACE1 fewer green/red autophagosomes and more red-only auto- protein (HA-HACE1)12. Empty vector (MSCV)-transfected lysosomes were detected in MG132-treated Hace1 WT NCM, Hace1 À / À MEFs were used as controls. HA-tagged HACE1

NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications 7 & 2014 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430

a HACE1-GFP p62 Overlay Puromycin Control

b Overlay HACE1-RFP GFP-LC3

NCM

3Tg LC

MG132 –/– Hace1

c Hace1+/+ Lc3Tg Hace1–/– Lc3Tg kDa GFP-LC3 40

40 Gapdh

MG 132 – + – +

d Overlay HACE1-GFP Lamp1 NCM

–/– MG132 Hace1

Figure 5 | HACE1 contributes to autophagic clearance of protein aggregates in a p62-dependent manner. (a) Fluorescence microscopy images showing HACE1 (green) co-aggregating with p62 (red) in the perinuclear space on puromycin treatment (8 h) in Hace1 À / À NCM transiently expressing HACE1-GFP. p62 was visualized by immunofluorescent staining. Arrows indicate co-localized puncta. (b) Representative confocal images showing HACE1 (red) localizes to LC3 þ puncta (green) induced by MG132 treatment (8 h) in Hace1 À / À GFP-Lc3Tg NCM cells transiently expressing HACE1-RFP. Arrows indicate co-localized puncta. (c) Primary NCMs from Hace1 þ / þ Lc3Tg and Hace1 À / À Lc3Tg newborn pups were treated with or without MG132 for 8 h and the cell lysates were subjected to western blot with anti-GFP antibodies. Gapdh was used as a loading control. (d) Fluorescence microscopy images showing co-localization of HACE1 (green) with Lamp1 (red) on MG132 treatment (8 h) in Hace1 À / À NCM cells transiently expressing HACE1-GFP. Lamp1 was visualized by immunofluorescent staining. Scale bar, 10 mm in all images. protein expression was readily detectable in HA-HACE1- electrophoresis (PAGE), coomassie blue staining detected transfected MEF but not in the empty vector (MSCV)-transfected several distinct bands in the anti-HA Co-IP pull-down products MEF (Fig. 7a). After resolved by SDS–polyacrylamide gel from HA-HACE1 MEF lysates but not in the control Hace1 À / À

8 NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications & 2014 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430 ARTICLE

a MG132 MG132+Baf A1 b 100 +/+

90 Hace1

P<0.0001 80 % Of yellow dot –/– 70 +/+ –/– Hace1 Hace1 Hace1 MG132 + + + + Baf A1 – + – +

c Hace1 –/– WT+Empty vector +WT-HACE1 +HACE1(C876S) MG132

d 100 P<0.001

90

80 % Of yellow dot

70 WT +Empty +Wt- +HACE1 vector HACE1 -C876S

Hace1–/–

Figure 6 | Hace1 is essential for the completion of autophagic flux. (a,b) Primary NCMs from WT and Hace1 À / À newborn pups and (c,d) WT and Hace1 À / À MEF, as well as Hace1 À / À MEF stably transfected with WT HACE1 and C876S-mutant HACE1, were transiently transfected with a tandem tagged mRFP-GFP-LC3 reporter (tfLC3) plasmid, followed by 8 h treatment with MG132 or MG132 plus Baf in the last 4 h, and were subjected to confocal microscopy. DMSO treatment only was used in the control group. (a,c) Representative confocal micrographs of images used for quantification of the percentage of yellow dots in b and d. At least 10 images from each group were used for the calculation; error bars represent s.e.m., P-value as indicated (one-way ANAVO). Scale bars, 10 mm. Autophagosomes show both GFP and mRFP-LC3 signals (yellow) while autolysosomes exhibit mRFP-L3 signals only (red). MSCV MEF (Fig. 7b). The anti-HA Co-IP pull-down products were proteins found in multiple HA-HACE1 purifications were further subjected to MS analysis. We performed a total of five submitted for statistical enrichment of gene ontology pathway experimental replicates for HA-HACE1-expressing MEF and four analysis using the Gene Ontology For Functional Analysis experimental replicates for control Hace1 À / À MSCV-MEF. A total software tools in ArrayTrack (ArrayTrackhttp://www.fda. of 173 different proteins were identified from all nine purifications gov/ScienceResearch/BioinformaticsTools/Arraytrack)37. Terms with 158 identified in any of the five HA-HACE1 purifications. To associated with ‘cytoskeletal binding and organization’, ‘the identify potential HACE1 interactors of high confidence, the unfolded protein response’ and ‘protein complex disassembly’ proteins were filtered by selecting proteins that were found in were found to be statistically enriched (Pr0.05, nZ5, E42, more than or equal to two HA-HACE1 purifications, and then were Fisher’s exact test; Fig. 7d). Among them, several cytoskeletal found to be at least twofold enriched, by spectral counts, in the HA- proteins like gelsolin38, Fhl1 (refs 39–41) and vinculin42,43 were HACE1 purification when compared with the control sample. This found to be the important mediators of cardiomyopathy. Western data reduction yielded 73 potential binding partners (Fig. 7c and blot analysis confirmed that p62, gelsolin and Unc-45b are among Supplementary Table 1). these binding partners of HACE1 (Fig. 7e). Moreover, when A gene ontology term enrichment analysis was performed to Co-IP was performed of the same MEF lysate with anti Unc-45b determine which molecular functions and biological processes antibody, HA-HACE1 was readily detectable (Fig. 7f). In were over-represented. Gene names corresponding to the 73 addition, when the HA-HACE1 MEF Co-IP pull-down complex

NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications 9 & 2014 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430

a c WT (HA-HACE1) KO (MSCV) g (MEF-HA-pull-down): Hace1 (heart tissue) Cald1 MSCV HA-HACE1 MSCV H-HACE1 IB: kDa Actn4 kDa Capza1 110 110 HA Hsp90b

Capza HACE1 40 Capzb Hsp60 60 Gapdh Tubb 2c p62 Hspa9 110 Cnn2

b HA-pull-down Input Unc-45b Uba52 kDa

60 Hsc70 110 HACE1 Hsc70 110 80 Gsn 110 60 Gelsolin Cald1 50

40

Coomassie blue staining d Molecular function h (MEF-HA-pull-down): 30 Unfolded protein binding (heart tissue) MSCV HA-HACE1 Cytoskeletal protein binding 20 Actin binding 110 M HA- MSCV HA- HACE1 HACE1

e HA-Pull-down f Unc45b-pull-down proteins IB: 50 + kDa kDa kDa

110 60 110 IB:Ub p62

HACE1 30 IB:HA 50 IP-HA- IP- Input HA HACE MSCV - HACE1 HA- MSCV 110 HACE1 110 Unc-45b IB:

IPInput IP Gelsolin IP Input IP Input HA-HACE1 MSCV MSCV HA-HACE1 Figure 7 | Identification of HACE1 interacting proteins by MS analysis of HA Co-IP. (a,b) Hace1 À / À MEF cells were stably transfected with HA-tagged WT HACE1 (HA-HACE1) and empty vector MSCV. (a) Immunoblot for HA shows stable expression of HA-tagged HACE1 in HA-HACE1 MEF cell lysate. (b) HA-HACE1 MEF and control MSCV MEF cell lysates were subjected to Co-IP with anti-HA antibody and coomassie blue staining shows a couple of distinct bands only detected in HA-HACE1 MEF IP pull-down as indicated by red arrows. (c–f) Identification of HACE1 interacting proteins by MS analysis of Co-IP products. (c) Hierarchical clustering (heat) map demonstrating the clustering of proteins based on their expression intensities in HA-HACE1 MEF and control MSCV MEF HA-IP pull-down products. To the left of the heat map, selected annotations found to be enriched in the protein clusters are shown. n ¼ 5 for HA-HACE1 group and n ¼ 4 for control MSCV group. (d) Pie chart shows enriched representative HACE1-interacting proteins according to molecular function by GO term. (e) Selected HACE1-interacting proteins identified by HA-IP MS were evaluated by western blot with the indicated antibodies (IB). HA-HACE1 denotes HA-IP pull-down from HA-HACE1 MEF; MSCV denotes HA-IP pull-down from control MSCV MEF; and input indicates HA-HACE1 MEF lysate before IP. (f) HA-HACE1 MEF and control MSCV MEF lysates were subjected to co-IP with an anti-Unc-45b antibody and the pull-down were analysed by western blot with an anti-HA antibody. (g,h) HA-IP pull-down products from HA-HACE1 MEF (HA-HACE1) and Hace1 KO MSCV MEF (MSCV) immortalized in HA beads were further subjected to Co-Co-IP with WT mouse heart lysates, and the pull-down products were analysed by western blot using the indicated antibodies. HACE1 and those proteins with which it interacts were still detectable (g). There were also increased ubiquitinated proteins in the Co-Co-IP heart tissues (h). were further bound with whole-heart lysate (Co-Co-IP) from Accumulation of Hace1 binding partners in Hace1 À / À sTAC HACE1 WT mice, HACE1 and its interaction proteins p62, heart. On the basis of above observations, we predicted that Unc-45b, Hsc70, gelsolin and caldesmon 1 were still detec- proteins interacting with HACE1 would accumulate in the hearts table (Fig. 7g). Furthermore, there were also increased of sTAC-treated Hace1 À / À mice. We tested whether HACE1 ubiquitinated proteins in the Co-Co-IPs from heart tissues deficiency results in accumulation of those proteins in the stres- (Fig. 7h). This indicates that HACE1 is binding to these proteins sed myocardium in vivo under haemodynamic stress. We in the heart in addition to MEF. observed significantly increased accumulation of many HACE1-

10 NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications & 2014 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430 ARTICLE interacting proteins, such as Fhl1, vimentin, Unc-45b, galdesmon domain E3 ligase family. The difference is instead of the C2 1, b-tubulin (Fig. 8a,d), gelsolin and vinculin (Fig. 8b,d) in the domain, HACE1 has an ANK repeat domain at its N terminal Hace1 À / À sTAC heart compared with Hace1 þ / þ controls. and the function of this domain in HACE1 has not been explored. Molecular chaperones Hsp70, Hsc70 and Hsp25 were also sig- We demonstrated here that HACE1, similar to SMURF1, is a dual nificantly accumulated in the Hace1 À / À sTAC heart (Fig. 8c,d). function E3 ligase that can promote degradation of selective cargo In addition, expression levels of control proteins that were not by at least two independent mechanisms that differ in their identified in our HACE1 Co-IP assay, like connexin 40 and requirement for E3 ligase activity. HACE1 participates in connexin 43, were not affected by Hace1 deficiency (Supplemen- proteasomal degradation via its HECT domain E3 ligase activity tary Fig. 7). Thus, HACE1 deficiency leads to the accumulation of and contributes to autophagic degradation via its ANK domain- its binding partners under haemodynamic stress in the heart. mediated protein–protein interactions. Remarkably, the role of HACE1 in cellular proteostasis only becomes apparent under stress. This may indicate that redundant Discussion pathways of protein degradation can complement for HACE1 Levine and colleagues recently demonstrated that the HECT deficiency in the basal state, and that these pathways are impaired domain E3 ligase SMURF1 promotes degradation of viral capsid or inadequate during cellular stress. Our data indicate that proteins independently of its E3 ligase activity but requiring its HACE1 deficiency impairs autophagy in the heart during membrane-binding C2 domain44. These authors proposed that haemodynamic stress. Besides catalysing the ubiquitination of the C2 domain of SMURF1 may participate in the delivery of active Rac1 (refs 13–15), the roles of HACE1 in mediating Golgi selective autophagic substrates to the nascent autophagosome. In traffic16, suppression of RAR signalling17 and the degradation of this manner, SMURF1 may participate in both proteasomal specific proteins may underlie this autophagy defect. degradation (via its E3 ligase activity) and autophagy (via its C2 Many HACE1-interacting proteins are associated with domain). Both HACE1 and SMURF1 belong to the HECT cytoskeletal function. A properly balanced cytoskeletal quality

a d 20 Hace1+/+ Hace1–/– 15 P<0.001 Sham sTAC Sham sTAC kDa 10 Fhl1 5 50 0 Fhl1 8 6 P<0.01 4 Vimentin 2

50 Vimentin 0 Cald1 110 4 P<0.001 Unc-45b 110 3 2

β-Tubulin Cald1 1 50 0 Gapdh 40 4 3 P<0.001 2 1 Unc-45b 0 +/+ –/– b Hace1 Hace1 8 Sham sTAC Sham sTAC 6 P<0.001 4

Gelsolin -Tubulin 2 80 β 0 2 Vinculin 110 P<0.05 40 1 Gapdh Gesolin 0 6 P<0.001 4 2 Vinculin 0 c Hace1+/+ Hace1–/– 8 6 ShamsTAC Sham sTAC P<0.001 4 80

Hsp70 2 Hsp70 0 4 Hsc70 3 P<0.01 80 2 Hsc70 1 Hsp25 0 20 8 40 6 P<0.01 Gapdh 4

Hsp25 2 0 ShamsTAC Sham sTAC Hace 1+/+ Hace1–/–

Figure 8 | Proteins interacting with HACE1 accumulate in Hace1 À / À sTAC heart. (a–c) Representative western blots and (d) quantification show increased accumulation of some of the HACE1-interacting proteins in the Hace1 À / À sTAC myocardium. n ¼ 3 mice per group, error bars represent s.e.m., P-value between KO and WT sTAC as indicated (one-way analysis of variance). Gapdh was used as a loading control.

NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications 11 & 2014 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430 control is essential for cellular and organismal structure beginning of the descending aorta and tied around a 27-gauge blunt needle, which and function45–47. The maintenance of cardiac cytoskeletal was subsequently removed. At the end of the procedure, the chest and skin were closed. The mice were kept on a heating pad until responsive to stimuli. homoeostasis becomes even more essential in the context of Sham-operated animals underwent the identical procedure, except that the aortic disease, where appropriate remodelling is necessary for survival. constriction was not placed. The integrity of the sTAC operation was confirmed by Cytoskeleton-specific chaperones and Ub ligases are necessary for measuring the systolic pressure gradient shortly after the operation, which the assembly and degradation of cytoskeletal proteins and generated a systematic increase in systolic pressure from sham- (78 mm Hg) to 45 sTAC (170 mm Hg) -operated heart. Owing to the high early mortality rate of the constitute the protein quality control system in the heart . Hace1 À / À mice post sTAC, the experiment was only carried out for 4 days. However, very little is known about the actual mechanisms responsible for regulating the turnover of the cardiac 46 Cardiac function evaluation. Cardiac function was evaluated prior to sTAC and cytoskeleton . A recent study suggested that specific Ub E3 at day 2 and day 4 after sTAC using echocardiography as described previously 2. In protein ligases regulate degradation pathways for actin (via brief, mice were sedated with 1% isoflurane/oxygen, and echocardiographic images TRIM32)46 and tubulin (via Parkin)47. There is also evidence that were recorded using an Acuson Sequoia C256 System (Siemens Medical Solutions) the co-chaperones Unc-45 and Hsp90 are required for myosin with a 15L8 transducer. Depth and frequency were set at 2 cm and 13 MHz, quality control4. In the present study, we found HACE1 respectively. The heart was first imaged in 2D mode in the parasternal short-axis view. From this view, an M-mode cursor was positioned perpendicular to the physically binding with Unc-45B both by IP and reverse IP interventricular septum and posterior wall of the Left ventricular (LV) at the level (Fig. 7e–f). When we subjected the Co-IP products to MS of the papillary muscles, and M-mode images were obtained for measurement of analysis; indeed, proteins associated with cytoskeleton binding wall thickness and chamber dimensions with the use of the leading-edge con- and organization and the unfolded protein response and protein vention adapted by the American Society of Echocardiography. LV end-diastolic dimension (LVEDD) and LV end-systolic dimension (LVESD) were measured; complex disassembly were enriched (Fig. 7c,d). This suggests that percentage of LV fractional shortening (%FS) was described as [(LVEDD-LVESD)/ HACE1 may form a complex with Unc-45b, Hsc70, Hsp90 as well (LVEDD) Â 100%]. as other chaperons and cytoskeletal binding and organization proteins to facilitate the proper organization and turnover of Histology, immunohistochemistry and TEM. For morphometry, hearts were cytoskeletal proteins. Thus, our data suggest that HACE1 is a arrested in diastole with 1 M KCl, fixed with neutral buffered 10% formalin stress inducible, selective safeguard for cytoskeleton homoeostasis solution (Sigma HT501128), embedded in paraffin and sectioned to a thickness of that is required for normal cardiac responses to haemodynamic 5 mm. Haematoxylin and eosin-stained sections were used for measurement of stress. heart morphology and cardiomyocyte cross-sectional area using SimplePCI version 6 image analysis software (Hamamatsu Corporation). For immunohistochemistry, In conclusion, we present herein for the first time that HACE1 samples were fixed with buffered 4% paraformaldehyde, and paraffin sections were is a key regulator of the Ub-selective autophagy pathway. performed with minimal antigen retrieval (0.3% Triton X-100) using the following Moreover HACE1’s functions in selective autophagy is indepen- antibodies: mono- and polyubiquitinylated conjugates (FK2) (Enzo Life Science, dent of its E3 Ub ligase activity, but via protein–protein PW8810, 1:100), p62 (Progen, GP62C, 1:100), GFP (Life Technologies, G10362, 1:200). Following overnight incubation with primary antibody, the sections were interactions mediated by its ANK domain. We propose that incubated with a matching biotinylated secondary antibody and detected by using HACE1 could be the missed master regulator in the control of VECTASTAIN ABC (peroxidase) system (Vector Labs). Negative controls were autophagic degradation of damaged organelle and proteins. Our performed for all immunostaining by omission of the primary antibody. For TEM in vivo studies demonstrate that HACE1 is a novel critical analysis, fresh isolated LV were minced into 1–2 mm3 cubes and fixed with 2% glutaraldehyde in 0.1 M cacodylate buffer for 1 h at room temperature. Finally, the regulator of cardiac hypertrophy and remodelling. Our findings tissue was cut into 70 nm sections with an ultramicrotome, placed on TEM grids that HACE1 is upregulated in cardiac tissue of heart failure and imaged with a FEI Tecnai TEM operated at 80 kV at the Pathology Laboratory patients both make it an attractive candidate as a potential of the Hospital for Sick Children according to standard procedures. diagnostic biomarker for heart failure and provide an insight into the mechanisms of cardiac remodelling. Immunoblotting. Whole-cell lysates from tissue and cell samples were prepared on ice with cell lysis buffer (Cell Signaling, 9803) containing a Complete cocktail of Methods proteases inhibitors (Roche Applied Science, 05892791001). Lysates were cleared Detection of HACE1 mRNA. The collection and use of human samples in this by centrifugation at 12,000 r.p.m. for 10 min. The supernatants were collected and study were approved by the Research Ethics Board of University Health Network protein concentrations determined using the Bio-Rad Bradford protein assay kit and informed consent was obtained from patients/family members. HACE1 mRNA (Bio-Rad). Twenty micrograms of protein lysate was separated by NuPAGE expression in human and mouse samples was detected by quantitative real-time SDS–PAGE Gel System (Life Technologies) and electrophoretically transferred to reverse transcriptase (RT)–PCR analyses using standard procedures. Isolation of polyvinylidene difluoride membranes (Invitrogen). Membranes were incubated total RNA was performed using the Trizol reagent (Invitrogen). cDNAs were overnight at 4 °C with antibodies reactive to the following proteins: ubiquitin synthesized from 1 mg total RNA with the SuperScript III First-Strand Synthesis (P4D1) (Cell Signaling, 3936, 1:1,000), p62 (Progen, GP62C, 1:1,000), LC3 (Novus, System for RT–PCR (Invitrogen) according to the manufacturer’s instructions. NB100-2331, 1:500), GFP (Life Technologies, A11121), HA (Monoclonal Antibody Quantitative RT–PCR was done using SYBR green assays (Roche Applied Science) Facility, Hospital for Sick Children, 12CA5, 1:1,000), Unc-45b (GenWay, with HACE1-specific primers and both HPRT and GAPDH as an endogenous EV0087,1:1,000), gelsolin (BD, 610412, 1:1,000), vimentin (R28) (Cell Signaling, control. Real-time PCR reactions were carried out in 384-well plates using a Roche- 3932, 1:1,000), Fhl1 (Novus, NB100-1461, 1:1,000) vinculin (Sigma, V4505, LightCycler 480 System. All primer sequences are listed in Supplementary Table 2. 1:1,000), b-tubulin (Sigma, T8328, 1:1,000), caldesmon (Abcam, ab32330, 1:1,000), Hsc70 (Abcam, ab2788, 1:1,000), Hsp70 (Stressgen, SPA810, 1:1,000), Hsp25 (Stressgen, SPA801, 1:1,000), connexin 40 (Life Technologies, 36-5000, 1:1,000) Mice strains and creation of pressure overload mouse model. Hace1-null mice and connexin 43 (Life Technologies, 71-0700, 1:1,000). Blots were incubated with with a mixed B6;129Ola genetic background were generated as reported pre- horseradish peroxidase-conjugated goat anti-mouse IgG (Bio-Rad, 170-5046, viously12, subsequently backcrossed to C57BL/6J mice (Jackson Laboratory) for 1:50,000), goat anti-rabbit IgG (Bio-Rad, 170-5047, 1:100,000), monoclonal mouse more than six generations and used for all the experiments reported in this study. anti-goat IgG (Jackson ImmunoResearch, 205-032-176, 1:50,000) or goat anti- GFP-LC3#53 transgenic mice21 were imported from RIKEN Bio-Resource Center guinea pig IgG (Jackson ImmunoResearch, 106-035-003, 1:100,000) and developed and were crossed with Hace1 À / À .mice to generate Hace1 À / À /GFP-LC3Tg or using ECL or ECL Plus Western Blotting Detection System (Amersham Hace1 þ / þ /GFP-LC3Tg mice. Mice were genotyped by PCR12 and maintained at Biosciences). To normalize signals to total protein, blot membranes were stripped the Animal Resources Centre of the University Health Network. All animal and re-probed with antibody against Gapdh (Cell Signaling, 2118, 1:2,000). experimental protocols were approved by the Animal Care and Use Committee of The quantification of signals was performed by densitometry of scanned the University Health Network and performed in accordance with the institutional autoradiographs with the aid of Quantity One software (version 4.6; Bio-Rad). guidelines. Proteasome activities were measured by the Chemicon Proteasome Activity Male Hace1 À / À and Hace1 þ / þ mice (15–17-week old) with body weight from Assay Kit (Millipore, APT280). In brief, freshly harvested heart samples were lysed 27 to 29 g were subjected to pressure overload by sTAC as previously described13. in lysis buffer (50 mM HEPES (pH 7.5), 5 mM EDTA, 150 mM NaCl and 1% In brief, mice were anaesthetized with ketamine (IP; 90 mg kg À 1) and Rompun (IP; Triton X-100) supplemented with 2 mM ATP and were cleared by centrifuge at 10 mg kg À 1). The chest was opened, and a horizontal skin incision was made at 15,000 r.p.m. for 20 min at 4 °C. Equal amounts of protein (determined by Bio-Rad the level of the two to three intercostal space. The start of the descending aorta was Bradford protein assay) were used for preparing assay mixtures in a 96-well identified right after the subclavian branch. A 7-0 silk suture was placed around the fluorometer plate and the remaining steps were conducted following the

12 NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications & 2014 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430 ARTICLE manufacturer’s instruction. Fluorescence data were collected using a Molecular GFP-LC3 reporter assay, cells were mounted with ProLong Gold antifade reagent Devices Spectra Max M5 plate reader using 380 nm excitation and 460 nm with 40,6-diamidino-2-phenylindole (Invitrogen, P-36931) and subjected to con- emission. focal examination on an Olympus FluoView 1,000 Laser Scanning Confocal Microscope. Co-localization efficiency of mRFP with GFP signals of tfLC3 puncta Culture of mouse neonatal ventricular cardiomyocytes. Cell culture of neonatal was measured using ImageJ software, and has been shown as the percentage of the mouse cardiomyocytes was prepared from newborn Hace1 À / À and WT mouse total number of mRFP puncta. hearts within 48 h of birth. In brief, after trimming, left ventricles were mechani- For HACE1 co-localization with p62 and Lamp1, an immunostaining step with 2 þ 2 þ anti-p62 (Progen, GP62C) and anti-Lamp1 antibody (C-20) (Santa Cruz Biotech) cally minced in Ca - and Mg -free Hank’s balanced salt solution (HBSS) on þ ice, and then subjected to stepwise enzymatic digestion with 0.15% trypsin (Invi- was added before confocal examining. For HA-HACE1 co-localization with Ub trogen) in disassociation solution (137 mmol l À 1 NaCl, 5.36 mmol l À 1 KCl, protein aggregates, HA-HACE1 was detected with an anti-HA mouse IgG1 À 1 À 1 À 1 monoclonal antibody prelabelled with the Zenon Alexa Fluor 488 Mouse IgG1 0.81 mmol l MgSO4, 5.55 mmol l dextrose, 0.44 mmol l KH2PO4, À 1 À 1 Labeling Kit (Life Technologies, Z-25002); Ub-conjugated proteins were labelled 0.34 mmol l Na2HPO47H2O and 20.06 mmol l HEPES, pH 7.4). Cells released after the first digestion were discarded, whereas cells from subsequent using an anti-FK2 antibody (Enzo Life Science, PW8810, 1:100), followed by digestion were transferred into GIBCO DMEM/F-12 medium (Invitrogen) sup- staining with Alexa Fluor 633 goat anti-mouse IgG (Life Technologies, plemented with 10% fetal bovine serum (FBS; Invitrogen) until all cardiac cells A21052,1:1,000). Fluorescence microscopy was carried out on a Nikon Eclipse were isolated (B5 times). The resulting mixture was centrifuged for 5 min at TE2000-U. 800 r.p.m., resuspended in DMEM/F-12 medium and preplated for 2 h to remove non-cardiomyocytes based on the observation that non-muscle cells attach to the substrata more rapidly. The cardiomyocytes were then collected and plated on Generation of HACE-expressing MEF and immunoprecipitation. Hace1 À / À laminin-coated culture plates at a density of 2 Â 106 cells ml À 1 in DMEM/F-12 MEF stably expressing an N-terminal HA-tagged full-length human HACE1 cDNA plus 10% FBS and 0.1 mM 5-bromodeoxyuridine (to inhibit growth of non-myo- (HA-HACE1) was generated using a retroviral system as described previously11.In cytes). Cells were incubated at 37 °C with 5% CO2 in a humidified atmosphere. A brief, HA-tagged HACE1cDNA was generated by subcloning the HACE1 coding confluent monolayer of spontaneously beating cardiomyocytes was formed within region from the IMAGE cDNA clone number 4838835 (ATCC; GenBank accession 2 days and was ready for downstream gene transfer and treatment. number BC034982) into mammalian expression vector pcDNA3-HA (Life Technologies). The HA-tagged Cys to Ser mutant (C876S) form of HACE1 was generated by Site directed mutagenesis using the QuickChange kit (Stratagene). Protein aggregate clearance assay. Protein aggregate clearance assays were HA-tagged HACE1 cDNA was subsequently subcloned into pMSCV-hygromycin performed with both neonatal NCMs and MEF cell culture. MEF cells were pre- vector (Clontech) followed by transfection into the BOSC23 ecotropic retroviral À / À pared from hace1 or littermate WT E13.5 mouse embryos and were grown in packaging cell line using calcium phosphate precipitation. Retrovirus-containing À 1 DMEM (Invitrogen) containing 10% FBS (Invitrogen), 10 units ml penicillin supernatants were collected 48 h after transfection and used to infect Hace1 À / À À 1 À / À (Invitrogen) and 0.1 mgml streptomycin (Invitrogen). Hace1 MEF stably MEF. To generate stably transfected MEF cells, infected cells were selected using expressing WT HACE1, C876S E3 ligase dead mutant (HACE1C876S) and empty 200 mgmlÀ 1 hygromycin (Life Technologies) for 48 h. 11 vector were generated using a retroviral system as described previously . The three The HACE1 immunoprecipitation (IP) analysis was performed using anti-HA À / À GFP-tagged HACE1 plasmids used for transient transfection into hace1 MEF Affinity Matrix (Roche Applied Science, 11 815 016 001); for Unc-45b IP, were from OriGene; pCMV-AC-HACE1-GFP (RG205602), HACE1DHECT and immunoprecipitation kit (Protein G) (Roche Applied Science, 11 719 836 001) with HACE1DAN were generated by gene synthesis and subcloned into pCMV-AC- goat polyclonal anti Unc-45b (GenWay, EV0087, 1:50) as IP antibody. Equal mGFP plasmid (OriGene). For protein aggregate clearance assay, MEF cells were amounts of cell lysate were immunoprecipitated with anti-HA or anti-Unc-45b and grown on Millicell EZ slides (Millipore) to about 70% confluence and were treated were subjected to SDS–PAGE and immunoblotted with the indicated antibodies. À 1 with 10 mgml puromycin (Sigma, P8833) for 4 h to induce ubiquitinated protein To confirm that HACE1 is also binding to the same proteins as in the heart, the aggregate formation, followed by extensive washing. Aggregate clearance was anti-HA Affinity Matrix that has been incubated with HA-HACE1 MEF or control followed for a further 8 h (total 12 h) incubation with fresh medium. Non-treated MSCV MEF lysates was further incubated with whole-heart lysate (Co-Co-IP) from MEFs were used as negative controls. After treatment, MEF were washed three HACE1 WT mice and equal amounts of eluate were subjected to immunoblotting times in PBS, followed by fixing in 4% paraformaldehyde and were subjected to with the indicated antibodies. fluorescence staining for Ub-conjugated protein with an anti-FK2 antibody (Enzo Life Science, PW8810, 1:100). In brief, fixed cells on the slides were first permeabilized for 2 min in 0.1% Triton-X 100, and then blocked for 1 h in PBS with MS analysis To map out the molecular interaction of HACE1, we examined the 3% normal goat serum and 1% BSA. Slides were next incubated overnight at 4 °C . anti-HA Co-IP products by MS analysis, essentially as previously described41,48.In with FK2 antibody, followed by staining with Alexa Fluor 633 goat anti-mouse IgG brief, samples were subjected to an in-solution tryptic digestion and then subjected (Life Technologies, A21052, 1:1,000) for 1 h. After being nuclei stained with to solid-phase extraction using OMIX C-18 pipette tips (Varian Inc.). MS Hoechst 33342 (Life Technologies, H1399), cells were mounted with ProLong Gold experiments were carried out on a LTQ linear ion-trap mass spectrometer (Thermo antifade reagent (Life Technologies, P-36934) and subjected to confocal Fisher Scientific) equipped with a nanoelectrospray source (Proxeon Biosystems, examination on an Olympus FluoView 1000 Laser Scanning Confocal Microscope. Denmark). Peptides were eluted from the microcapillary columns using a 2-h The total fluorescence intensity from the FK2 staining was measured using gradient as described41. Spectra were identified by converting Raw MS files to m/z Olympus Fluoview FV1000 viewer software. XML using ReAdW and searched by a locally installed version of X!Tandem against a mouse IPI (International Protein Index; http://www.ebi.ac.uk/IPI) protein Small interference RNA and p62 immunoprecipitation. Knockdown of p62 in sequence database created on August 2010. To estimate and minimize our false- Hace1 þ / þ and Hace1 À / À NCMs was achieved using ON-TARGET plus positive rate, the protein sequence database also contained every IPI protein SMARTpool mouse Sqstm1 (p62) siRNAs (Thermo/Dhamacon l-047628-01); ON- sequence in its reversed amino-acid orientation (target-decoy strategy) as recently TARGET plus non-targeting siRNA pool (D-001810-01) were used as control. described41,48. In this study we set the value of total reverse spectra to total forward DharmaFECT 1 Transfection Reagent (T-2001) was used to transfect siRNAs in a spectra to 0.5%, resulting in a low number of decoy sequences in the final output. final concentration of 20 nM to NCM cells and knockdown of p62 mRNA was Only peptides matching these criteria were accepted to generate the final list of confirmed by RT–PCR using p62 gene-specific primers. Forty-eight hours after identified proteins, with further criteria set to select only those proteins identified knockdown, cells were subjected to protein aggregate clearance assay as illustrated with a minimum of two unique peptides. in the above section. p62 immunoprecipitation was performed by using Pierce Classic Magnetic IP/Co-IP kit (Thermo Scientific, 88804) with guinea pig poly- clonal anti-p62 (Progen, GP62C 1:100) as IP antibody; and the pull-down products Statistical analysis. For all multiple comparisons of normally distributed data, were analysed by western blot with anti-mono and polyubiquitin-conjugated one-way analysis of variance followed by Bonferroni’s multiple comparison post- antibody (FK2) (Enzo Life Sciences, BML-PW01510, 1:1,000). tests were used, while for comparison of two groups of normally distributed data as shown in Fig. 1b, unpaired two-tailed Student’s t-tests were used. Survival rates were analysed by the Kaplan–Meier method, and difference between groups was Plasmid transfection and microscopy. Transient transfection of plasmid con- tested with the log-rank w2-test. Statistical analyses were conducted using Graph- structs into both primary mouse NCM and MEF was carried out by using the Pad Prism 5 software. All values are presented as means±s.e.m.; n refers to the Transfection Reagent Xfect (Clontech, 631318) or a Neon Transfection System sample size. P 0.05 was considered significant. (Life Technologies) according to standard protocols. HACE1 expression plasmids o pCMV6-AC-HACE1-GFP and pCMV6-AC-HACE1-RFP were from OriGene. The tandem tagged mRFP-GFP-LC3 reporter (tfLC3) plasmid was from Addgene. Forty-eight hours after transfection, cells were treated with either 10 mgmlÀ 1 References puromycin or 1 mM MG132 with or without 100 nM Baf for the indicated time, 1. de Couto, G., Ouzounian, M. & Liu, P. P. Early detection of myocardial followed by washing three times in PBS. Cells were then fixed in 4% paraf- dysfunction and heart failure. Nat. Rev. Cardiol. 7, 334–344 (2010). ormaldehyde, permeabilized for 2 min in 0.1% Triton-X 100 and then blocked for 2. Kuba, K. et al. Impaired heart contractility in Apelin gene-deficient mice 1 h in PBS with 3% normal goat serum and 1% BSA. For the tandem tagged mRFP- associated with aging and pressure overload. Circ. Res. 101, 32–42 (2007).

NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications 13 & 2014 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4430

3. Sun, M. et al. Tumor necrosis factor-alpha mediates cardiac remodeling 35. Taylor, J. M., Brody, K. M. & Lockhart, P. J. Parkin co-regulated gene is and ventricular dysfunction after pressure overload state. Circulation 115, involved in aggresome formation and autophagy in response to proteasomal 1398–1407 (2007). impairment. Exp. Cell Res. 318, 2059–2070 (2012). 4. Kassiri, Z. et al. Combination of tumor necrosis factor-alpha ablation and 36. Kimura, S., Noda, T. & Yoshimori, T. Dissection of the autophagosome matrix metalloproteinase inhibition prevents heart failure after pressure maturation process by a novel reporter protein, tandem fluorescent-tagged LC3. overload in tissue inhibitor of metalloproteinase-3 knock-out mice. Circ. Res. Autophagy 3, 452–460 (2007). 97, 380–390 (2005). 37. Sun, H., Fang, H., Chen, T., Perkins, R. & Tong, W. GOFFA: gene ontology for 5. Nakai, A. et al. The role of autophagy in cardiomyocytes in the basal state and functional analysis - a FDA gene ontology tool for analysis of genomic and in response to hemodynamic stress. Nat. Med. 13, 619–624 (2007). proteomic data. BMC Bioinformatics 7(Suppl 2): S23 (2006). 6. Su, H. & Wang, X. The ubiquitin-proteasome system in cardiac proteinopathy: 38. Li, G-H. et al. Gelsolin regulates cardiac remodeling after myocardial infarction a quality control perspective. Cardiovasc. Res. 85, 253–262 (2010). through DNase I-mediated apoptosis. Circ. Res. 104, 896–904 (2009). 7. Rothermel, B. A. & Hill, J. A. Autophagy in load-induced heart disease. Circ. 39. Friedrich, F. W. et al. Evidence for FHL1 as a novel disease gene for isolated Res. 103, 1363–1369 (2008). hypertrophic cardiomyopathy. Hum. Mol. Genet. 21, 3237–3254 (2012). 8. Rothermel, B. A. & Hill, J. A. The heart of autophagy: deconstructing cardiac 40. Gossios, T. D., Lopes, L. R. & Elliott, P. M. Left ventricular hypertrophy proteotoxicity. Autophagy 4, 932–935 (2008). caused by a novel nonsense mutation in FHL1. Eur. J. Med. Genet. 56, 251–255 9. Zhu, H., Rothermel, B. A. & Hill, J. A. Autophagy in load-induced heart disease. (2013). Methods Enzymol. 453, 343–363 (2009). 41. Sharma, P., Shathasivam, T., Ignatchenko, V., Kislinger, T. & Gramolini, A. O. 10. Wang, Z. V., Rothermel, B. A. & Hill, J. A. Autophagy in hypertensive heart Identification of an FHL1 protein complex containing ACTN1, ACTN4, and disease. J. Biol. Chem. 285, 8509–8514 (2010). PDLIM1 using affinity purifications and MS-based protein-protein interaction 11. Anglesio, M. S. et al. Differential expression of a novel ankyrin containing E3 analysis. Mol. Biosyst. 7, 1185–1196 (2011). ubiquitin-protein ligase, Hace1, in sporadic Wilms’ tumor versus normal 42. Tangney, J. R. et al. Novel role for vinculin in ventricular myocyte mechanics kidney. Hum. Mol. Genet. 13, 2061–2074 (2004). and dysfunction. Biophys. J. 104, 1623–1633 (2013). 12. Zhang, L.-Y. et al. The E3 ligase HACE1 is a critical chromosome 6q21 tumor 43. Zemljic-Harpf, A. E. et al. Cardiac-myocyte-specific excision of the vinculin suppressor involved in multiple cancers. Nat. Med. 13, 1060–1069 (2007). gene disrupts cellular junctions, causing sudden death or dilated 13. Torrino, S. et al. The E3 ubiquitin-ligase HACE1 catalyzes the ubiquitylation of cardiomyopathy. Mol. Cell. Biol. 27, 7522–7237 (2007). active Rac1. Dev. Cell 21, 959–965 (2011). 44. Orvedahl, A. et al. Image-based genome-wide siRNA screen identifies selective 14. Castillo-Lluva, S., Tan, C. T., Daugaard, M., Sorensen, P. H. & Malliri, A. The autophagy factors. Nature 480, 113–117 (2011). tumour suppressor HACE1 controls cell migration by regulating Rac1 45. Willis, M. S., Schisler, J. C., Portbury, A. L. & Patterson, C. Build it up-tear it degradation. Oncogene 32, 1735–1742 (2013). down: protein quality control in the cardiac sarcomere. Cardiovasc. Res. 81, 15. Daugaard, M. et al. Hace1 controls ROS generation of vertebrate 439–448 (2009). Rac1-dependent NADPH oxidase complexes. Nat. Commun. 17, 2180 (2013). 46. Lundin, V. F., Leroux, M. R. & Stirling, P. C. Quality control of cytoskeletal 16. Tang, D. et al. The ubiquitin ligase HACE1 regulates Golgi membrane proteins and human disease. Trends Biochem. Sci. 35, 288–297 (2010). dynamics during the cell cycle. Nat. Commun. 2, 501–511 (2011). 47. Kim, J., Lo¨we, T. & Hoppe, T. Protein quality control gets muscle into shape. 17. Zhao, J., Zhang, Z., Vucetic, Z., Soprano, K. J. & Soprano, D. R. HACE1: a novel Trends Cell. Biol. 18, 264–272 (2008). repressor of RAR transcriptional activity. J. Cell. Biochem. 107, 482–493 (2009). 48. Ryan, T. et al. Identification of novel ryanodine receptor 1 (RyR1) protein 18. Tannous, P. et al. Intracellular protein aggregation is a proximal trigger of interaction with calcium homeostasis endoplasmic reticulum protein (CHERP). cardiomyocyte autophagy. Circulation 117, 3070–3078 (2008). J. Biol. Chem. 286, 17060–17068 (2011). 19. Zhu, H. et al. Cardiac autophagy is a maladaptive response to hemodynamic stress. J. Clin. Invest. 117, 1782–1793 (2007). Acknowledgements 20. Rothermel, B. A. et al. Differential activation of stress-response signaling in load-induced cardiac hypertrophy and failure. Physiol. Genomics 23, 18–27 (2005). We thank Youan Liu for technical help; N. Mizushima for permission to use the 21. Ahluwalia, N. et al. Prognostic value of multiple emerging biomarkers in GFP-LC3#53 transgenic mice; T. Yoshimori for ptfLC3 plasmids. This work was cardiovascular risk prediction in patients with stable cardiovascular disease. supported by grants from the Canadian Institutes of Health Research (CIHR 43865 and Atherosclerosis 228, 478–484 (2013). MT13392) and Heart and Stroke Foundation of Canada (H&S 452478) to P.P.L. and 22. Ghosh, N. & Haddad, H. Atrial natriuretic peptides in heart failure: Ontario Ministry of Research and Innovation Global Leadership Round In Genomics 2 pathophysiological significance, diagnostic and prognostic value. Can. J. And Life Sciences (GL -CBD3P3) to P.P.L. and A.O.G. Physiol. Pharmacol. 89, 587–591 (2011). 23. Kotby, A. A. et al. Atrial natriuretic peptide as a marker of heart failure in Author contributions children with left ventricular volume overload. J. Paediatr. Child Health 49, L.Z. co-designed the study, analysed the data, wrote the manuscript and together with 43–47 (2013). X.C. performed most of the experimentation; F.D. performed animal surgery; M.P.N. 24. Komatsu, M. et al. Homeostatic levels of p62 control cytoplasmic inclusion performed some earlier experiments; A.D.S. helped with confocal and editing of the body formation in autophagy-deficient mice. Cell 131, 1149–1163 (2007). manuscripts; M.M. performed the quantitative PCR assay and helped with Immuno- 25. Mizushima, N., Yoshimori, T. & Levine, B. Methods in mammalian autophagy histochemistry; P.S. and A.O.G. performed the mass spectrometry proteomic research. Cell 140, 313–326 (2010). experiments and analysis; J.W. and R.-K.L. performed echocardiographic measurement 26. Mizushima, N., Yamamoto, A., Matsui, M., Yoshimori, T. & Ohsumi, Y. in vivo and analysis; J.M.P., J.H.B. and P.H.S. provided ideas, reagents and data exchanges for analysis of autophagy in response to nutrient starvation using transgenic mice the project; P.P.L. supervised and co-designed the study; L.Z., J.H.B. and P.P.L. wrote the expressing a fluorescent autophagosome marker. Mol. Biol. Cell 15, 1101–1111 (2004). manuscript. All authors discussed the results and commented on the manuscript. 27. Szeto, J. et al. ALIS are stress-induced protein storage compartments for substrates of proteasome and autophagy. Autophagy 2, 189–199 (2006). 28. Huang, J. et al. Antibacterial autophagy occurs at PI(3)P-enriched domains of the Additional information endoplasmic reticulum and requires Rab1 GTPase. Autophagy 7, 17–26 (2011). Supplementary Information accompanies this paper at http://www.nature.com/ 29. Sussman, M. A. et al. Altered focal adhesion regulation correlates with naturecommunications cardiomyopathy in mice expressing constitutively active rac1. J. Clin. Invest. 105, 875–886 (2000). Competing financial interests: The authors declare no competing financial interests. 30. Matsumoto, G., Wada, K., Okuno, M., Kurosawa, M. & Nukina, N. Serine 403 phosphorylation of p62/SQSTM1 regulates selective autophagic clearance of Reprints and permission information is available online at http://npg.nature.com/ ubiquitinated proteins. Mol. Cell 44, 279–289 (2011). reprintsandpermissions/ 31. Bjørkøy, G. et al. p62/SQSTM1 forms protein aggregates degraded by How to cite this article: Zhang, L. et al. HACE1-dependent protein degradation autophagy and has a protective effect on huntingtin-induced cell death. J. Cell provides cardiac protection in response to haemodynamic stress. Nat. Commun. 5:3430 Biol. 171, 603–614 (2005). 32. Ding, W-X. et al. Linking of autophagy to ubiquitin-proteasome system is doi: 10.1038/ncomms4430 (2014). important for the regulation of endoplasmic reticulum stress and cell viability. Am. J. Pathol. 171, 513–524 (2007). This work is licensed under a Creative Commons Attribution- 33. Wu, W.-K. et al. Induction of autophagy by proteasome inhibitor is associated NonCommercial-NoDerivs 3.0 Unported License. The images or other with proliferative arrest in colon cancer cells. Biochem. Biophys. Res. Commun. third party material in this article are included in the article’s Creative Commons license, 374, 258–263 (2008). unless indicated otherwise in the credit line; if the material is not included under the 34. Ja¨nen, S. B., Chaachouay, H. & Richter-Landsberg, C. Autophagy is activated by Creative Commons license, users will need to obtain permission from the license holder proteasomal inhibition and involved in aggresome clearance in to reproduce the material. To view a copy of this license, visit http://creativecommons. culturedastrocytes. Glia 58, 1766–1774 (2010). org/licenses/by-nc-nd/3.0/

14 NATURE COMMUNICATIONS | 5:3430 | DOI: 10.1038/ncomms4430 | www.nature.com/naturecommunications & 2014 Macmillan Publishers Limited. All rights reserved.